Collaborative robot

Abstract

A robot (1) includes at least one articulated arm having a base (3) for components (3, 4-1, ..., 4-5, 5), an end effector (5), and a plurality of links (4-1, ..., 4-6), wherein each link (4-1, ..., 4-6) is movably connected to two other components (3, 4-1, ..., 4-6, 5) via corresponding joints (7-1, ..., 7-6), at least one sensor (10, 9) for detecting an external force acting on any one of the components (3, 4-1, ..., 4-6, 5), and a controller (6) for controlling the movement of the joints (7-1, ..., 7-6) to move the end effector (5) along a pre-programmed path (12). In the event that the sensor (9, 10) detects an external force, the controller (6) is adapted to employ a first release strategy to escape from the external force, and is adapted to evaluate whether the first strategy is successful; if unsuccessful, it is adapted to employ a second release strategy.

Description

TECHNICAL FIELD

[0001] The present invention relates to industrial robots for collaborative applications. BACKGROUND

[0002] Collaborative robots are designed to be used in the vicinity of people so that contact between the robot and the people cannot always be avoided or can even be necessary to accomplish a given task. Such contact can be constrained or unconstrained. Contact is called unconstrained if the involved body part moves freely away from the contact, while it is called constrained if the involved body part cannot move freely away from the contact. Furthermore, contact can be transient or quasi-static. Transient means that dynamic forces dominate and the robot impacts the body part at a perceptible speed. Transient contact can be unconstrained or constrained. Quasi-static means that driving forces dominate while dynamic forces can be neglected.

[0003] It is important that contact between the people and the robot can occur without risk of injury to the people. This can be achieved by limiting the speed of the arms of the robot so that the energy transferred in a possible contact with a person is safely below a potentially harmful level, and in case contact is detected, stopping the movement of the robot or possibly modifying the robot in a way that releases the contact. SUMMARY

[0004] The present invention relates to a robot adapted to release such contact when it occurs, while minimizing the discomfort of the people.

[0005] To this end, the invention provides a robot comprising at least one articulated arm having a base for the members, an end effector and a plurality of links, wherein each link is movably connected to two other of said members by a respective joint, at least one sensor for detecting an external force acting on any of said members, and a controller for controlling the movement of the connected members relative to each other in order to move the end effector along a preprogrammed path, characterized in that in case the sensor detects an external force, i.e. a force caused by an object external to the system and in particular not by the weight and inertia of the robot itself, the controller is adapted to employ a first release strategy for escaping from the external force along a first release path, to evaluate whether the first strategy was successful, and if not, to employ a second release strategy along a second release path different from the first release path. In this way, even some malfunctions should prevent the movement along the first release path, unintentional contact between the robot and the people can be released with minimal risk of discomfort and minimal possibility of requiring intervention by another person.

[0006] The optimal strategy for releasing the contact can depend on the properties of the environment in which the robot operates, for example on the properties of the surface against which the robot can press a body part of a person. As clamping between a given surface and the robot is only possible if a part of the robot is sufficiently close to the surface, it is advantageous to take the current pose of the robot into account when selecting a release strategy.

[0007] When an external force is detected that indicates that a contact has occurred, one practical release strategy is to control the robot so as to move the end effector along the path it followed before the contact but in the opposite direction. The advantage of this strategy is that, as long as the objects in the robot's environment do not move, the path is certainly free of obstacles and the computational effort for controlling the release movement is small, so that the release movement can be started with minimal delay.

[0008] However, the release movement according to the above strategy can not be the most efficient movement to actually release the contact between the robot and the person. In order to release the contact along the shortest possible path and thus also in the shortest possible time, another strategy is to detect the direction of the external force acting on one of the members of the robot and to control the displacement of said one member in the detected direction. Again, this strategy can not always be optimal; if the controller is provided with data about its environment, it can recognize that a movement in the direction of the contact force would lead to the robot contacting some obstacle, which would make the first-mentioned strategy appear preferable. If the controller is not provided with such data and a contact with an obstacle occurs, a movement away from the contact in the direction of the contact force can again put the person in danger. Obviously, in this case, it is useful to change the release strategy to move the robot in a way that releases the person quickly and permanently.

[0009] If the person touched by the robot has moved into a position that blocks the path, or if a technical malfunction prevents the backward movement of one of the joints of the robot, it is not possible to move the robot back along the incident path. In this case, as another release strategy, the controller can suitably select one of the joints of the robot and set the relative movement of the two members connected by said joint in the direction that yields to said load. If controlling the movement of this joint is not sufficient to eliminate the external force, for example because the selected joint happens to be the joint affected by the malfunction and cannot move despite being controlled, or because its range of displacement is not sufficient, another joint can be selected and the process can be repeated for this other joint.

[0010] If the controller is adapted to compare the external on different joints, it is best to select the joint whose external load is highest first, because this is also the joint for which the smallest amount of displacement is likely to release the contact.

[0011] The load can be a force transmitted from one link to another via a connecting joint, a torque acting on the joint, or a torque weighted by the moment of inertia of the member of the robot arm distal to the joint. If the joint is a rotational joint, the load can also be the above-mentioned torque or a component of the torque in the direction of the joint axis.

[0012] Whenever it is determined that the strategy of the robot for releasing the contact is not successful, one possible option is a so-called protective stop, in which not only the robot is stopped, but also the drive power is removed, in order to prevent any further dangerous occurrences.

[0013] The protective stop can be an appropriate strategy, in particular in the above-mentioned case where the release movement of the robot is blocked by an obstacle.

[0014] At the boundaries of the working range of the robot, there are points which the end effector can only reach by assuming a singular pose, in which the links of the robot connected by rotatable joints are arranged in a straight line. In this singular pose, an external force acting along the straight line does not generate a torque in the connecting joints, which means that the force which the robot can exert to a person in this singular pose cannot be controlled. Therefore, the controller should be adapted to prohibit the robot arm from assuming the singular pose.

[0015] The invention also relates to a method of controlling a robot, comprising the steps of:

[0016] - detecting whether an external force is acting on the robot; and, in the affirmative case,

[0017] - employing a first release strategy for escaping from the external force,

[0018] - evaluating whether the first strategy is successful, and if not, employing a second release strategy,

[0019] and a data carrier storing computer readable instructions which, if read and executed by a computer, cause the computer to operate as a controller in the above-mentioned robot or to perform the above-mentioned method. BRIEF DESCRIPTION OF DRAWINGS

[0020] Further features and advantages of the present invention will become apparent from the following description of an embodiment thereof, with reference to the drawings.

[0021] Figure 1 is a schematic view of a robot according to the invention;

[0022] Figure 2 a first release strategy of the robot is shown; and

[0023] Figure 3 a second release strategy is shown.

[0024] Figure 1The robot 1 and its installed working environment are schematically shown. DETAILED DESCRIPTION

[0025] The robot 1 has a conventionally designed articulated arm comprising a base 3, an end effector 5 and a plurality of links 4-1, 4-2,... connected to each other by revolute or prismatic joints, to the base 3 and to the end effector 5, and a controller 6 for controlling the movement of the arm, i.e. the movement of the end effector 5 relative to the base 3. For the sake of simplicity, only the revolute joints 7-1 to 7-6 are shown in Figure 1 The joints 7-1, 7-4 and 7-6 have an axis of rotation extending in the longitudinal direction of the adjacent links; the joints 7-2, 7-3 and 7-5 have an axis extending in the transverse direction.

[0026] The human body part 8 shown is clamped between the end effector 5 and the surface 11 of the working environment, which results in an external force acting on the end effector. According to a first embodiment, a sensor 10 for detecting the external force vector is provided in the end effector 5 itself.

[0027] According to an alternative embodiment, each joint 7-1 to 7-6 has a sensor 9 associated therewith for detecting the torque in the direction of its respective axis of rotation. The sensor can be a force-sensitive element, for example a piezoelectric element connected to the two links meeting in the joint, or it can be formed by circuitry for measuring the current flowing in the motor driving the relative rotation of said links, which current also represents the torque to which the motor is subjected. The torque measured by each such sensor can be considered as the sum of two contributions. The first contribution is a function of the weight and the pose of the robot members distal of the joint and of possible inertial forces acting on these members, the second contribution is the effect of the contact with the body part 8. Since the weight, the speed and the pose of all these distal components are known to the controller, the first contribution can be estimated. If this calculated contribution is significantly different from the torque detected by the sensor 9, it can be concluded that an external force is acting on the robot 1. From the measured torque in each joint and the calculated first contribution, the second contribution can be estimated, and from the estimated second contribution in each joint and the position of the end effector 5 relative to the joint axes, the external force vector can be calculated.

[0028] According to a third embodiment, two sensors 9, 10 are provided. In this case, if there is a discrepancy between the external force detected by the sensor 10 and the external force calculated from the data of the sensor 9, it can be determined that the external force is not acting on the end effector, but on a neighbouring member of the robot. The implications of this finding for possible release strategies will be discussed later.

[0029] In Figure 2In the middle, it is assumed that the end effector 5 comes into contact with the body part 8 by glancingly approaching the surface 11 along the path 12. While in many cases it would be appropriate that the end effector 5 after detecting an external force indicating contact with a person should move back along the same path 12 but in the opposite direction, this can not be the case in this situation. If the end effector 5 has already scraped the person's skin on its way to the shown position, it should not repeat the scraping on its way back. As mentioned above, the controller 6 avoids the risk by estimating the direction of the external force exerted by the body part 8 onto the end effector 5 and then controls the robot to move the end effector in that force direction. If the person's skin is still under shear stress from the end effector's advance, this direction can approximately coincide with the path 12; on the other hand, if the skin has slipped off and is not under shear stress but under pressure, then the path 13 of the releasing movement will approximately coincide with the surface normal of the body part 8. Even if the person unconsciously tries to withdraw the body part 8, the resulting external force will be detected and will cause the controller to move the end effector 5 together with the withdrawn body part 8, thus releasing the end effector 5 while avoiding a scratch.

[0030] In an advanced embodiment of the invention, the controller 6 is provided with data about the coordinates of surfaces 11 in the robot's working environment. Based on this information and the past and current pose of the robot, the controller 6 judges which surface the robot is approaching and at which angle. If the approach angle is close to the surface's normal when an external force indicating contact is detected, the robot can be controlled to release the contact by simply reversing its movement direction and thereby moving back along its path of arrival away from the surface, without detecting the direction of the external force. On the other hand, when the angle is large enough to risk a scrape, it is worth releasing the contact by backing up in the external force direction.

[0031] Figure 3 A situation is shown in which another surface 14 restricts the mobility of the robot 1. Due to the presence of this surface 14, only by moving the end effector 5 more or less along the path 12 can the shown pose of the robot be reached. Here reference is made to the first embodiment described above. Figure 2 The described releasing movement would bring the joints 7-5 of the robot close to the surface 14. In the first embodiment described above, no sensor detects this contact; all the controller 6 will notice that the external force detected by the sensor 10 does not decrease as it should and possibly that the position sensors at the joints 7-1 to 7-6 notice that the angles of the joints 7-1 and 7-3 do not change as expected if the robot 1 executes the commands from the controller 6 correctly. Thus, after a short time, the controller 6 judges that the first releasing strategy of withdrawing the end effector 5 from the surface 11 along the direction of the external force is not successful. Depending on the power of the motors driving the members of the robot 1, a few milliseconds or tens of milliseconds are sufficient for this judgment.

[0032] If the first release strategy is found to fail, the controller 6 switches to a second release strategy. This second strategy can be to retract the end effector 5 along the path 12 it came from.

[0033] Alternatively, the controller employs a strategy of moving the robot 1 joint by joint: According to the first embodiment, the controller determines which joint can be rotated to release the external force and in which direction, based on the known pose of the robot 1 and the force vector provided by the sensor 10. Trying to turn joint 7-2, all robot members distal to joint 7-2 can be expected to release the external force by turning joint 7-2 counterclockwise, but this cannot be done because surface 14 blocks joint 7-5. The same is true for joint 7-3. When the controller 6 selects joint 7-5, turning link 4-5, joint 7-6 and end effector 5 counterclockwise, the contact is successfully released.

[0034] According to the second embodiment, when the external force acting on the end effector is first detected, the controller 6 has already determined from the external torques at joints 7-2, 7-3 and 7-5 in which direction these torques should be rotated in order to release the contact. In joints 7-2, 7-3, the external torques (more precisely, the components of the external torques parallel to the axes of rotation of the joints) are largest, the controller 6 first selects one of these and controls it to rotate counterclockwise. Normally, this should immediately release the contact. In this case, it does not, because surface 14 blocks the rotation. The controller 6 switches to joint 7-5 upon detecting that controlling joint 7-2, 7-3 to rotate CCW does not release the torque detected in joint 7-5, and successfully releases the contact by turning it counterclockwise.

[0035] If the controller selects the joint to be rotated based on the torques weighted by the moments of inertia of the members distal to the joint, the first joint to be selected can be joint 7-5, because the only member distal to it is the end effector 5, and the moment of inertia can be small. Although the angle through which joint 7-5 has to be rotated in order to release the contact is larger than the angle through which a link at a larger distance has to be rotated, the low inertia allows joint 7-5 to be rotated faster, so that rotating joint 7-5 can be the fastest way to release the contact.

[0036] It should be noted that the invention can not only deal with the end effector 5, but also with other movable members of the robot that are in contact with the body part 8. In fact, if some intermediate link of the robot arm is in contact with the body part 8, this intermediate link and all members distal to it can be considered as the above-mentioned end effector 5, and the above-mentioned strategies can be applied to it.

[0037] Reference signs

[0038] 1 robot

[0039] 3 base

[0040] 4-1,... link

[0041] 5 end effector

[0042] 6 controller

[0043] 7-1,... joint

[0044] 8 body part

[0045] 9 sensor

[0046] 10 sensor

[0047] 11 surface

[0048] 12 path

[0049] 13 path

[0050] 14 surface

Claims

1. A robot (1) comprising at least one articulated arm having a base (3) for a member (3, 4-1,..., 4-6, 5), an end effector (5), and a plurality of links (4-1,..., 4-6), wherein each link (4-1,..., 4-6) is movably connected to two other members of the members (3, 4-1,..., 4-6, 5) by a respective joint (7-1,..., 7-6), at least one sensor (9, 10) for detecting an external force acting on any one of the members (3, 4-1,..., 4-6, 5), and a controller (6) configured to control movement of the respective joints (7-1,..., 7-6) in order to move the end effector (5) along a pre-programmed path, characterized in that in case the sensor (9, 10) detects an external force, the controller (6) is adapted to: employ a first release strategy for escaping from the external force, evaluate whether the first release strategy was successful and, when the first release strategy was not successful, employ a second release strategy for escaping from the external force, and prohibit the robot arm from assuming a single pose corresponding to the plurality of links being arranged in a straight line, wherein in the first release strategy or the second release strategy, the controller is further configured to select one of the respective joints and to control the one of the respective joints to rotate based on an external torque acting on the one of the respective joints.

2. The robot according to claim 1, wherein in the first release strategy or the second release strategy, the controller (6) is adapted to control the robot arm to move the end effector (5) in reverse along the pre-programmed path (12).

3. The robot according to claim 1, wherein the at least one sensor (9, 10) is adapted to detect a direction of the external force, and in the first release strategy or the second release strategy, the controller (6) is adapted to control displacement of the any one of the members (3, 4-1,..., 4-6, 5) on which the external force acts in the detected direction (13).

4. The robot according to any one of claims 1-3, wherein in the first release strategy or the second release strategy, the controller (6) is further adapted to select one of the respective joints (7-1,..., 7-6) and to control the one of the respective joints to move in yield to a direction of the detected external force.

5. The robot according to claim 4, wherein the controller is adapted to detect, by the sensor, more than one external force acting on one or more of the respective joints and to select the one of the respective joints on which the external force is highest.

6. The robot according to claim 4, wherein the controller is adapted to select another of the respective joints if the external force fails to be reduced.

7. The robot according to claim 5, wherein the controller is adapted to select another of the respective joints if the external force fails to be reduced.

8. An industrial system comprising a robot: at least one articulated arm comprising a base, an end effector, and a plurality of links, wherein each link of the plurality of links is movably connected to two other members by a respective joint, at least one sensor configured to detect an external force acting on any of the members, and a controller configured to control movement of the respective joints in order to move the end effector along a pre-programmed path, wherein the controller is configured to perform a method comprising the steps of: - detecting whether an external force is acting on the robot, and in the affirmative, - adopting a first release strategy for escaping from the external force, - evaluating whether the first release strategy is successful, and when the first release strategy is not successful, adopting a second release strategy for escaping from the external force, and - prohibiting the robot arm from assuming a single pose corresponding to the plurality of links being arranged in a straight line, wherein in the first release strategy or the second release strategy, the controller is further configured to select one of the respective joints and to control the one of the respective joints to rotate based on an external torque acting on the respective joint.

9. A data carrier storing computer readable instructions which, when read and executed by a computer, cause the computer to operate as a controller in a robot according to any of claims 1-8.

Images